Primary malignant brain tumors, like glioblastoma (GBM), remain universally fatal. Like most neoplasms, they develop through the acquisition of multiple genetic alterations that lead to a heterogeneous deregulation of cell signaling pathways. Despite this complexity, recent advances in gene expression technology have been successful at providing more accurate prognostic information to patients. Still, they have little impact on patient care because they do not alter treatment choice. Innovative approaches pioneered by our group, however, offer an opportunity to identify more elaborate structure in the patterns of gene expression in these tumors by extrapolating the findings obtained with defined cell culture manipulations in vitro, such as activation of a given cell signaling pathway, to the complexity of human cancers in vivo. The resulting "gene signatures" can be thought of as a "fingerprint" shared between experimental cell cultures and patient tumors. To the degree to which they are shared, we believe these signatures have the potential to be used as guides for directing the use of targeted therapeutic agents to treat human cancers. In support of this hypothesis, we have recently shown that these gene signatures accurately predict oncogenic pathway activation and response to targeted therapeutics in various murine and human tumors. The advantage of targeting therapy to susceptible tumors is well illustrated by the examples of trastuzumab for HER2-expressing breast cancer and imatinib for Philadelphia chromosome chronic myeloid leukemia. Similarly, support for the basic concept that genetic analysis can inform targeted therapy in GBM has recently been provided in two retrospective studies - one demonstrating that O6-methylguanine-DNA methyltransferase promoter methylation can inform the use of temozolomide chemotherapy in GBM and the other identifying a significant association between clinical response to epidermal growth factor receptor (EGFR) inhibitors in patients and tumors that co-express PTEN and EGFRvIII. These studies provide evidence that molecular analysis could be used to select patients with GBM that are more likely to respond to a given therapy. These observations combined with the dismal results of studies using various single agents in GBM, suggests that the complexity and heterogeneity of GBM will need to be matched with an equally complex therapeutic combination. This is not much different than in other biologic systems, for example AIDS, leukemia, or bacterial infections, where the potency of combinatorial therapy has been evident in many early and dramatic treatment successes. Therefore, our OVERALL GOAL in this proposal is to enhance the efficacy of targeted combinatorial therapeutics for patients with brain tumors. PUBLIC HEALTH RELEVANCE: Brain tumors remain the most common cause of cancer death among children and account for more deaths in adults than melanoma, and treatment for these tumors represents the most expensive medical therapy per quality-adjusted life-year saved currently provided in the United States. Innovative analysis of genetic "fingerprints" in these and other tumors may allow therapy to be enhanced by matching specific susceptibilities of the tumor to targeted therapeutic agents. This proposal tests this "personalized" medicine approach in human tumors grown in mice as a prelude to human clinical studies.